aderose <aderose@tripology.com> writes:
> Starting with a database where analyze has never been run I get worse
> performance after running it -- is there something I'm missing?
Well, not basing such a sweeping statement on a single query example
would be a good start ;-). This particular plan might have got a little
worse but I'm sure some further investigation would show other cases
that got better.
If you want to nudge it back towards the indexscan plan, reducing
random_page_cost a bit would probably do the trick. But that's
definitely not a parameter you want to adjust on the basis of only
one test case.
regards, tom lane